Most cold email campaigns reply below the average
Across 81 campaigns and 236,927 contacted leads, the pooled reply rate is 2.55 percent. The median campaign replied at 2.12 percent, and 50 of 81 - 62 percent - came in below the pooled figure. Fourteen campaigns produced half of all replies. When an agency quotes you an average, this is the distribution hiding behind it.
Every campaign, plotted
One bar per campaign, sorted worst to best. Two dashed lines: the pooled average everyone quotes, and the median campaign that half the set fell below.
Best quartile 4.41%, worst quartile 0.81%. A 5.4 times spread between the top and bottom quarter of the same operation, same infrastructure, same year. Anyone quoting a single number for this channel is describing one point on that range and calling it the range.
Test a threshold against the real set
If a vendor promises you a reply rate, this is how often that promise was met across our own book. Set the number they quoted and see how many of 81 real campaigns cleared it.
How many campaigns cleared it
With JavaScript unavailable, the figures shown are for a 2.50 percent threshold.
Read this the right way round. A campaign clearing a threshold does not mean the threshold is achievable on demand - these are outcomes measured after the fact, on lists and copy that varied enormously. It tells you how often it happened, not how often it can be promised.
Where the gap comes from
Splitting by campaign size locates it precisely. Below 5,000 contacted leads the pooled average and the median campaign agree almost exactly. Above 5,000 they come apart completely.
| Leads contacted | Campaigns | Contacted | Pooled | Median campaign | Gap |
|---|---|---|---|---|---|
| 500 to 2,000 | 33 | 40,191 | 1.99% | 2.00% | 1x |
| 2,000 to 5,000 | 37 | 123,302 | 2.69% | 2.56% | 1.1x |
| 5,000 and above | 11 | 73,434 | 2.62% | 0.98% | 2.7x |
In the 5,000 and above band the pooled rate is 2.62 percent while the median campaign managed 0.98 percent - a 2.7 times gap, from 11 campaigns and 73,434 contacted leads. Two large campaigns carried that band. The rest of it did poorly.
Why this matters more than the headline number. Large campaigns are exactly the ones a buyer is quoted on, because they are where the volume is. They are also where the average is least representative of what actually happens. The smaller the campaign, the more honest the average.
What this means when you are buying
Ask which number they are quoting
Pooled average and median are both honest and they answer different questions. Pooled tells you what the whole book did. Median tells you what a typical campaign did. On our own set they differ by 1.2 times overall and by 2.7 times on large campaigns. An agency quoting only the first is not lying, but it is not describing your likely experience either.
Ask for the spread, not the centre
The useful question is not "what is your average reply rate" - it is "what did your worst quarter of campaigns do". Ours did 0.81 percent. Our best quarter did 4.41 percent. If a vendor cannot answer that, they either do not measure per campaign or would rather not say.
Assume you land in the middle, then check the maths still works
Planning against the pooled average means planning against an outcome 50 of our 81 campaigns did not reach. Plan against the median instead, and use the cost per opportunity engine to see whether the deal still works at that rate. If it only works at the average, it does not work.
What we are not claiming
That our numbers are the industry's. This is one agency's book over five months. A different operation with a tighter niche would show a different spread - possibly a much better one. The finding here is about the shape of the distribution, which we would expect to survive across operations, not about the level.
Method and limitations
The sample
Every campaign in our sending platform across 7 client programmes, pulled 12 August 2026: 132 campaigns in total, of which 81 had contacted at least 500 leads and are included here. Together they contacted 236,927 leads and received 6,038 unique replies.
The definition
Reply rate is unique replies divided by leads contacted, per campaign - not per email sent, and not counting a lead twice for replying twice. This is the same definition published on cold email benchmarks, so the two pages are directly comparable.
The 500-lead floor, and why
Campaigns below 500 contacted leads are excluded from every figure. A campaign that contacted three leads and got one reply is a 33 percent reply rate and pure noise; including it would inflate the spread and make the finding look stronger than it is. Fifty-one of the 132 campaigns fall below the floor and appear nowhere on this page.
A counting error we caught
Campaigns are stored per workspace and the platform's switch-workspace call can return success while leaving the session in the previous workspace. Our first pull produced 108 rows with only 95 distinct campaign ids - 13 campaigns counted twice under two different clients, and one whole programme missing. Every figure here comes from the verified re-pull, in which each workspace was confirmed to return a campaign set that did not overlap one already collected. Had we not checked, the per-programme numbers would have attributed one client's campaigns to another.
What this does not measure
- Reply quality. A reply is any reply, including "no thanks" and "wrong person". Of the 6,038 replies here, 204 were tagged interested, but that tag measures how hard an inbox was triaged rather than how good the reply was, so we do not build on it.
- Meetings or revenue. Those are on benchmarks and the pricing page, from a separate programme with its own denominators.
- Cause. This shows that the spread exists, not why. List quality, offer, copy, seniority and timing all vary across these campaigns and this page does not separate them.
- Send pacing. We tried to cut by daily new-lead cap and could not: every campaign above the floor sits in a single pacing bucket, so there is no contrast to measure. The cut is reported as impossible rather than presented with a spurious result.
Bounces ran at 1.33 percent of emails sent and unsubscribes at 0.043 percent of leads contacted across the same set. Our standards are on methodology and editorial standards.
Related
If you are comparing quotes, the number worth asking for is the worst quarter, not the average. We are happy to show you ours.
Apply to work with us